Switching the Cofactor Specificity of an Imine Reductase

Article abstract of DOI:10.1002/cctc.201701194

In the last years, imine reductases (IREDs) have gained importance for the formation of chiral amines by catalyzing asymmetric reductions of imines and chemo- and stereoselective reductive aminations. However, all characterized members of this steadily gr

Full text of DOI:10.1002/cctc.201701194

DOI: 10.1002/cctc.201701194
Full Papers
Switching the Cofactor Specificity of an Imine Reductase
Niels Borlinghaus and Bettina M. Nestl*[a]
In the last years, imine reductases (IREDs) have gained impor-
tance for the formation of chiral amines by catalyzing asym-
metric reductions of imines and chemo- and stereoselective re-
ductive aminations. However, all characterized members of this
steadily growing enzyme family demonstrated strict preference
for NADPH, which entails reduced possibilities for efficient co-
factor regenerations and limitations in the construction of
whole cell systems. To alter the cofactor specificity from
NADPH to NADH, we applied the “Cofactor Specificity Rever-
sal— Structural Analysis and Library Design” (CSR-SALAD) muta-
genesis strategy and enlarged the mutant library by further
amino acid replacements. This engineering approach has been
shown to result in IRED variants with up to 2900-fold improved
NADH/NADPH specificity and completely recovered activity in
the reduction of 2-methyl pyrroline (2MP).
Introduction
Chiral amines are powerful building blocks of high significance,
which are used for the synthesis of a variety of industrially rele-
vant chemicals.[1] Over the last few years, different methodolo-
gies have been developed for their synthesis. In this light,
transaminases,[2] lipases,[3] ammonia lyases,[4] amine dehydro-
genases,[5] monoamine oxidases,[6] and imine reductases
(IRED)[7] are efficient biocatalysts for preparing chiral amines in
high stereoselectivity. Although the majority of these enzymes
generate only primary amines, IREDs can catalyze imine reduc-
tions and reductive aminations with direct access also to sec-
ondary and tertiary amines featuring excellent enantiopurity
(Scheme 1).[7–10] A broad range of amines was produced by
using IREDs that were applied as isolated enzymes, in multi-
enzyme cascades, and whole cell approaches.[11–14]
pletion and allow the removal of oxidized cofactor. The devel-
opment of various NAD(P)H cofactor regeneration processes
has been subject to intensive studies.[15–20] In most regenera-
tion schemes for nicotinamide cofactors, the desired enzymatic
reduction step is coupled to another enzymatic reaction. Sev-
eral enzymatic methods have been described, such as the re-
duction of NAD+ with the formate dehydrogenase (FDH) from
Candida boidini,[21,22] the phosphite dehydrogenase (PTDH)
from Pseudomonas stutzeri,[23] or the soluble hydrogenase from
Ralstonia eutropha H16.[24] These systems distinguish them-
selves because they are cost-saving substrates, have excellent
atom efficiency, and do not hinder the product workup after
reaction.[25]
In contrast to the regeneration of NADH, the accessibility of
NADPH regenerating enzymes is rare and the available systems
show quite low applicability.[15] Different approaches were pro-
posed to overcome these limitations, including engineered for-
mate dehydrogenase and phosphite dehydrogenase variants
with switched cofactor specificity[27–31] and whole cell systems
using the natural cofactor regeneration by cellular metabolism.
However, the allocation of NADPH is limited to about
0.1 UmgÀ1 cell dry weight, and might be not sufficient for en-
zymes with high activity.[15] Furthermore, there have been
many reported successes of enzyme redesign for altered cofac-
tor specificity utilizing directed evolution, site-directed muta-
genesis, or (semi)-random saturation mutagenesis.[32,33]
Scheme 1. General scheme of IRED-catalyzed reactions. Imines and iminium
ions are reduced by IRED by using NADPH as cofactor to afford amines. The
imine substrate (shown in blue) can be directly reduced or formed through
reaction of a carbonyl group with an amine.
To enable economic biotransformations on the industrial
scale, in situ cofactor regeneration of the expensive nicotina-
mide cofactor is required. Thus, only a catalytic amount of
NAD(P)H is needed and the cost can be notably reduced. Fur-
thermore, cofactor regeneration can drive the reaction to com-
Recently, the group of Hçhne reported the generation of an
IRED variant by rational design demonstrating a three-fold in-
creased activity for NADH with 10% activity recovery.[34] To ad-
dress this important challenge of improving the specificity and
activity for NADH, we altered the cofactor specificity of the R-
selective IRED from Myxococcus stipitatus (R-IRED_Ms) by using
the recently described online tool “Cofactor Specificity Rever-
sal— Structural Analysis and Library Design” (CSR-SALAD), de-
veloped in the lab of Frances H. Arnold. CSR-SALAD is a semi-
rational structure-guided approach that aims at reversing the
nicotinamide cofactor specificity of enzymes with Rossman
[a] N. Borlinghaus, Dr. B. M. Nestl
Institute of Biochemistry and Technical Biochemistry, Chair of Technical Bio-
chemistry
University of Stuttgart
Allmandring 31, 70569 Stuttgart (Germany)
E-mail: bettina.nestl@itb.uni-stuttgart.de
Supporting information for this article can be found under:
https://doi.org/10.1002/cctc.201701194.
ChemCatChem 2018, 10, 183 –187
183
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 2. Selected target residues of R-IRED_Ms for switching the cofactor
specificity. Residues in the phosphate coordinating pocket are highlighted in
magenta. Further residues are shown in light orange.
The decrease in absorbance at 340 nm was monitored in the
conversion of 2-methylpyrroline (2MP) to 2-methylpyrrolidine
(2MPD) by using cell-free lysates. In the first round of muta-
genesis, we were able to identify four variants with improved
specificity. Interestingly, all of them carry a tyrosine at posi-
tion 33, followed by polar (N, T) or even charged residues (D,
E, K) at positions 34 and 37 (Figure S2 in the Supporting Infor-
mation). Best results were obtained by the double variant
R33Y/T34E (V1) with no substitution at position K37. The ne-
cessity of the combined modifications was tested by generat-
ing the R33Y single mutant, which showed 21-fold lower spe-
cificity than V1 (Figure S5 in the Supporting Information).
To test whether additional substitutions at position 37 could
contribute to the properties of the double variant, we supple-
mented the library, performing a site-saturation mutagenesis
by using V1 as a template. The substitution of lysine by ala-
nine or arginine had a marked effect on either the specificity
(YEA, V2) or activity (YER, V3). The data are given in Figure 3
and Figure S3 (in the Supporting Information).
Figure 1. Rossman fold NADPH binding pocket of (R)-selective IRED from
Myxococcus stipitatus. Specificity determining regions are shown in magenta
(2’-phosphate coordinating) and light orange (not 2’-phosphate coordinat-
ing). The 3D model was generated by using the SWISS-MODEL server (tem-
plate PDB code 3ZHB[26]).
fold binding pockets as in the case of IREDs (Figure 1).[35] This
approach includes the simultaneous mutagenesis of phosphate
coordinating residues, which generally determine the cofactor
specificity. Compared with a combined saturation mutagenesis,
the suggested mutant library is strongly reduced and, thus,
does not require high throughput screening. Furthermore,
CSR-SALAD predicts additional positions in the cofactor bind-
ing pocket to recover the activity subsequently with a handful
of site-saturation mutagenesis approaches. This promising tool
and the first insights from the group of Hçhne inspired us to
switch the cofactor specificity of R-IRED_Ms while ensuring re-
covery of the enzymatic activity. Herein, we report the step-
wise mutagenesis in the cofactor binding pocket of R-IRED_Ms
by utilizing CSR-SALAD.
Both variants were used as templates for the site-saturation
mutagenesis at position 32, which was predicted by CSR-
SALAD to mainly influence the activity. Although no increase in
activity could be observed, remarkable improvements in spe-
cificity were noticed by introducing the variant N32E (V4 and
Results and Discussion
Owing to the fact that the crystal structure of R-IRED Ms has
not been determined yet, we used the structure of the closely
related R-selective IRED from Streptomyces kanamyceticus (R-
IRED_Sk, PDB code 3ZHB, 42% identity, Supporting Information
Figure S1) for CSR-SALAD analysis. Cahn et al. demonstrated in
previous work the robustness of CSR-SALAD including one pro-
tein without a crystal structure.[35] Recommended mutations
for R-IRED_Sk at seven positions were transferred onto R-IRED_
Ms. Four positions were located around the 2’-phosphate
moiety of NADPH and the other amino acids were positioned
opposite to the phosphate binding pocket (Figure 2).
We generated the first mutant library by changing three of
the four phosphate coordinating residues (R33, T34, K37) si-
multaneously against the suggested amino acids. We used
three degenerated codons (HVC, RMK, and DMK) coding for
different numbers of amino acids, resulting in a reduced
mutant library of 432 members (Table S2 in the Supporting In-
formation). For the screening of the mutant library, we devel-
oped a deep-well plate-based NAD(P)H consumption assay.
Figure 3. Turnover frequencies with NADH (dark gray) and NADH/NADPH
specificities (gray) of generated R-IRED_Ms variants compared with the wild-
type enzyme (WT).
ChemCatChem 2018, 10, 183 –187
www.chemcatchem.org
184
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
V5). Variant V4 demonstrated the highest specificity; however,
the overall activity was distinctly reduced. After three rounds
of mutagenesis and screening, we were able to identify two
promising variants (V2 and V5) with about five times increased
activities and over 500-fold improved specificities towards
NADH. Further, we performed site-saturation mutagenesis at
positions 67, 71, and 75. We thus focused our engineering on
V5 and subsequently transferred the most promising substitu-
tions to V2. Saturating position L75 did not lead to improved
variants. In contrast to L75, improved variants were obtained
by saturating positions L67 and T71. Substituting L67 with iso-
leucine led to a slightly increased specificity and activity (V6).
Through mutagenesis of T71, three variants with remarkable
improvements were identified. Interestingly, all of them
showed hydrophobic replacements (Figure S4 in the Support-
ing Information), with T71V featuring the highest activity (V7).
We then combined L67I and T71V (V8) and observed strong
synergistic effects, reaching a 47-fold increased NADH activity
and a 900-times improved specificity compared with the wild-
type (Figure 3, Table 2). Remarkably, we achieved a completely
recovered activity with a turnover frequency (TOF) of 94 minÀ1
comparable to the wild-type with 91 minÀ1. Similar observa-
tions were made when we transferred L67I and T71V to variant
2 (V9), resulting in a TOF of 84 minÀ1. To further increase the
specificity of V9, we introduced mutation N32E (V10). This var-
iant showed a 2900-times improved specificity compared with
the wild-type. However, a reduction of activity with NADH was
detected (TOF=38 minÀ1).
Next, we performed biotransformations by using 2-MP as
the substrate to evaluate product formations and selectivities
(Table 2). Biocatalytic reactions were stopped after 45 min,
shortly before the wild-type enzyme reached full conversion
with NADPH. At the same time, the tested variants formed
only 46% (V8) and 11 % (V10) of product (2MPD) if NADPH
was used. The opposite development was noticed if NADH
was applied. In contrast to the wild-type (14% product forma-
tion), variants V8 and V10 showed excellent conversions of
>98% and 94%, respectively. Remarkably, the high enantiose-
lectivity of the wild-type enzyme with >98% ee (R) was un-
changed for both variants.
For a better understanding of the structural changes in the
cofactor binding pocket of V8 and V10, we calculated possible
arrangements in silico with YASARA.[36] We ran 0.5 ns molecular
dynamics (MD) refinements by using energy-minimized homol-
ogy models as the starting structures. The comparison of the
resulting structures enabled us to presume how individual resi-
dues influence the cofactor stability (Figure 4). The most influ-
ential mutation R33Y is found in all initial variants. One impor-
tant impact of this mutation might be the destabilization of
NADPH, as the positively charged arginine is described to be
the favored interaction partner of the negatively charged 2’-
phosphate.[35] However, this only occurs when further phos-
phate coordinating residues in positions 32, 34, and 37 were
replaced. In this context, the introduction of glutamic acid at
positions 32 or 34 might cause a reduced acceptance of
NADPH by electrostatic repulsion.
The kinetic parameters of V8 and V10 were determined to
further validate the influence of these mutations in the asym-
metric reduction of 2-MP by using NADPH and NADH as cofac-
tors (Table 1 and Figure S6 in the Supporting Information).
Confirming the observed activities and specificities, both var-
iants showed improved catalytic efficiencies if NADH was used.
The kcat value of V8 with NADH was even 1.7-fold higher com-
pared with the wild-type with its natural cofactor.
Position 37 seems to be also involved in the stabilization of
NADH and the residue at this position is the only discriminator
between variants V8 and V10. We assumed that arginine at
this position (V8) is able to stabilize NADH better than alanine
(V10). However, K37R should reduce the repulsion of NADPH
owing to its positive charge. This might explain why variants
with K37R show a lower specificity for NADH than variants
with other mutations at this position, such as K37A, K37M, or
K37Q (Figure S3 in the Supporting Information).
Table 1. Kinetic parameters of variants 8 and 10 compared with the wild-type (WT) enzyme by using 2-methylpyrroline (2MP) as the substrate.
NADPH
KI [mm]
NADH
KI [mm]
KM [mm]
kcat [sÀ1
]
kcat/KM [minÀ1/mmÀ1
]
KM [mm]
kcat [sÀ1
]
kcat/KM [minÀ1/mmÀ1
]
WT
V8
V10
1.0Æ0.1
12Æ2
2.3Æ0.1
20Æ2
11 Æ2
16Æ4
134Æ16
0.7Æ0.1
0.2Æ0.03
23Æ1
0.09Æ0.003
3.9Æ0.04
1.5Æ0.2
12Æ1
21Æ0.2
29Æ6
0.2Æ0.01
24Æ1
0.14Æ0.01
0.02Æ0.002
9.8Æ0.3
11 Æ2
6.8Æ0.9
7.9Æ1.6
Table 2. Activities and specificities of variants 8 and 10 compared with the wild-type enzyme by using 2-methylpyrroline (2MP) as the substrate.
NADPH
Conversion[b]
[%]
NADH
Conversion[b]
[%]
Specificity
NADH/NADPH[a]
Fold change
[minÀ1/minÀ1
TON[a]
[minÀ1
ee[b]
[%]
TON[a]
[minÀ1
ee[b]
[%]
]
]
]
from wild-type
WT
V8
V10
91Æ2
>98
>98 (R)
>98 (R)
>98 (R)
2.0Æ0.07
94Æ10
14Æ1.7
>98
94Æ3
>98 (R)
>98 (R)
>98 (R)
0.022Æ0.001
20Æ2
–
4.8Æ0.2
46Æ2
900Æ100
0.6Æ0.02
11 Æ1
38Æ5
64Æ8
2900Æ400
[a] 10 mm 2-methylpyrroline, 0.4 mm NAD(P)H, pH 7.0, 258C. [b] 8 mm 2-methylpyrroline, 10 mm NAD(P)H, 2 mm enzyme, 45 min, 258C, 180 rpm.
ChemCatChem 2018, 10, 183 –187
www.chemcatchem.org
185
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
Figure 4. 3D model of the wild-type NADPH binding pocket in R-IRED_Ms (WT) and the cofactor binding pockets of V8 and V10 with bound NADH instead of
NADPH. The shown arrangements were calculated with Yasara[36] running 0.5 ns MD refinements by using energy minimized homology models as starting
structures. Selected and mutated regions are shown in magenta (2’-phosphate coordinating) and light orange (not 2’-phosphate coordinating). Calculated dis-
tances between the 2’-phosphate moiety and residues R33, T34, and K37 (WT) are 2.6–3.0 Š. The distance between the 6’-carbon of adenine and tyrosine at
position 33 (V8, V10) is 4.4 Š. Distances between the 2’- and 3’-hydroxyl groups of NAD(H) and glutamic acid at position 32 (V8, V10) are 2.6–3.5 Š. The dis-
tance between the 3’-hydroxyl group of NAD(H) and arginine at position 37 (V8) is 3.0 Š.
tein expression was induced with the addition of arabinose (final
concentration 0.02%). Then, the cultures were incubated for about
20 h at 258C. Cells were harvested and then lysed with a high-pres-
sure homogenizer. Protein purification was performed with cobalt
His-Trap columns (His-GraviTrap-TALON, Healthcare) using buffer A
(50 mm potassium phosphate buffer pH 7.0, 300 mm KCl) for bind-
ing, buffer B (50 mm potassium phosphate buffer pH 7.0, 300 mm
KCl, 5 mm imidazole) for washing, and buffer C (50 mm potassium
phosphate buffer pH 7.0, 300 mm KCl, 500 mm imidazole) for elut-
ing the enzyme. Afterwards, the buffer was changed by dialysis
(two times, 2 h in 5 L, 50 mm potassium phosphate buffer pH 7.0,
MWCO=6–8 kDa). Purity and size was verified by SDS-PAGE (Fig-
ure S10 in the Supporting Information). The protein concentration
was determined by using the BCA Protein Assay Kit (Thermo Scien-
tific).
Furthermore, we assume that the exact positioning of the
cofactor in the binding pocket is essential for the stabilization
of NADH. In this regard, R33Y appears to be particularly impor-
tant as the tyrosine residue allows the rearrangement of the
cofactor through stacking interactions to adenine. To allocate
the NADH cofactor in a novel binding mode, alterations in the
native adenine binding pocket by mutations L67I and T71V
were required.
The elimination of a hydrogen bridge through substitution
of T71 against valine shall result in a reduced affinity of the co-
factor in its native conformation. Consequently, MD refine-
ments in silico verified that the cofactor is slightly rotated and
shifted (Figure 4). We hypothesize that this reorientation facili-
tates the establishment of stronger binding modes so that the
induced NADH specificity is accompanied by high activity.
Library creation and screening
Conclusions
For the simultaneous exchange of selected amino acids at three
positions (R33, T34, K37), degenerated codons (Table S1 in the Sup-
porting Information) suggested by CSR-SALAD were used. We used
the primer to integrate the mutations via Gibson assembly. For the
site-saturation mutagenesis approaches, we used the QuikChan-
geTM protocol from Stratagene by using a primer with the degener-
ated NNK codon (N=A, C, G, T; K=T, G) at the corresponding posi-
tion. Chemical competent E. coli JW5510 cells were transformed
with the modified DNA (60 s, 428C), spread on LB plates
(34 mgmLÀ1 chloramphenicol), and grown at 378C overnight to
obtain between 100 and 400 chloramphenicol-resistant colonies.
The presented engineering procedure using CSR-SALAD serves
as a powerful tool for introducing enzyme specificity towards a
non-native cofactor. We identified NADH-dependent IRED var-
iants by altering the cofactor binding pocket of the (R)-selec-
tive IRED from Myxococcus stipitatus. The mutagenesis of the
2’-phosphate binding pocket resulted in considerable improve-
ments regarding the NADH/NADPH specificity. Additional sub-
stitutions of L67 and T71 resulted in variants with reversed
specificity and recovered activity.
For the screening, single colonies were picked to inoculate the pre-
culture by using deep well plates (34 mgmLÀ1 chloramphenicol).
After incubation overnight at 258C, the preculture was used to in-
oculate the expression culture (34 mgmLÀ1 chloramphenicol, 0.02%
arabinose). The next day, the cells were harvested and stored at
À808C overnight. Afterwards, the pellets were resuspended in lysis
buffer (50 mm potassium phosphate buffer pH 7.0, 750 mgmLÀ1 ly-
sozyme, and 10 mgmLÀ1 DNAse I) and incubated for 1 h at 378C.
The suspension was centrifuged and the IRED-containing superna-
tant was used for the NAD(P)H assay (50 mm potassium phosphate
buffer pH 7.0, 10 mm 2-methylpyrroline, 0.4 mm NAD(P)H). The
NAD(P)H consumption was detected for 30 min at 340 nm in the
Experimental Section
Cloning, protein expression, and purification
The gene of R-IRED_Ms was cloned into the pBAD33 plasmid using
a Gibson Assembly and fused with an N-terminal his6-tag. For the
expression, we used E. coli JW5510 cells harboring the vector and
let them grow as preculture overnight at 378C. Next, we inoculat-
ed terrific broth (TB) culture medium containing 34 mgmLÀ1 chlor-
amphenicol with 0.25% preculture. After a 2–3 h incubation at
378C and reaching an optical density of OD600 =0.8–1.0, the pro-
ChemCatChem 2018, 10, 183 –187
www.chemcatchem.org
186
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Papers
[7] J. Mangas-Sanchez, S. P. France, S. L. Montgomery, G. A. Aleku, H. Man,
M. Sharma, J. I. Ramsden, G. Grogan, N. J. Turner, Curr. Opin. Chem. Biol.
2017, 37, 19–25.
[8] P. N. Scheller, M. Lenz, S. C. Hammer, B. Hauer, B. M. Nestl, ChemCatCh-
em 2015, 7, 3239–3242.
plate reader. For the activity calculation, we determined the rela-
tion between A340 and the NAD(P)H concentration to get the molar
attenuation coefficient of eNADH =3.2632 A340 mmÀ1 and eNADPH
=
3.4069 A340 mmÀ1
.
[9] H. Kohls, F. Steffen-Munsberg, M. Hçhne, Curr. Opin. Chem. Biol. 2014,
19, 180–192.
Enzyme activity assay
[10] G. Grogan, N. J. Turner, Chem. Eur. J. 2016, 22, 1900–1907.
[11] P. Matzel, M. Gand, M. Hçhne, Green Chem. 2017, 19, 385–389.
[12] S. P. France, S. Hussain, A. M. Hill, L. J. Hepworth, R. M. Howard, K. R.
Mulholland, S. L. Flitsch, N. J. Turner, ACS Catal. 2016, 6, 3753–3759.
[13] F. Leipold, S. Hussain, D. Ghislieri, N. J. Turner, ChemCatChem 2013, 5,
3505–3508.
[14] Z. Maugeri, D. Rother, J. Biotechnol. 2017, 258, 167–170.
[15] W. A. van der Donk, H. Zhao, Curr. Opin. Biotechnol. 2003, 14, 421–426.
[16] R. Wichmann, D. Vasic-Racki, in Technology Transfer in Biotechnology. Ad-
vances in Biochemical Engineering, Vol. 92, (Ed.: U. Kragl), Springer,
Berlin, 2005, pp. 225–260.
For the detection of the enzyme activity, we used defined concen-
trations of purified enzyme (0.2–20 mm) and incubated them to-
gether with 10 mm 2-methylpyrroline and 0.4 mm NAD(P)H in the
plate reader at room temperature. The cofactor consumption was
measured for 30 min at 340 nm. For kinetic studies, we used the
same protocol with varying concentrations of the substrate.
Biotransformation and sample preparation for the GC
[17] C. Rodriguez, I. Lavandera, V. Gotor, Curr. Org. Chem. 2012, 16, 2525–
2541.
[18] W. Liu, P. Wang, Biotechnol. Adv. 2007, 25, 369–384.
[19] A. Weckbecker, W. Hummel, Microb. Enzymes Biotransform. 2005, 17,
225–238.
[20] C.-H. Wong, G. M. Whitesides, J. Am. Chem. Soc. 1981, 103, 4890–4899.
[21] A. S. Bommarius, K. Drauz, W. Hummel, M. R. Kula, C. Wandrey, Biocatal.
Biotransform. 1994, 10, 37–47.
[22] Z. Shaked, G. M. Whitesides, J. Am. Chem. Soc. 1980, 102, 7104–7105.
[23] J. M. Vrtis, A. K. White, W. W. Metcalf, W. A. van der Donk, Angew. Chem.
Int. Ed. 2002, 41, 3257–3259; Angew. Chem. 2002, 114, 3391–3393.
[24] J. Ratzka, L. Lauterbach, O. Lenz, M. B. Ansorge-Schumacher, Biocatal. Bi-
otransform. 2011, 29, 246–252.
[25] L. Lauterbach, O. Lenz, K. A. Vincent, FEBS J. 2013, 280, 3058–3068.
[26] M. Rodríguez-Mata, A. Frank, E. Wells, F. Leipold, N. J. Turner, S. Hart,
J. P. Turkenburg, G. Grogan, ChemBioChem 2013, 14, 1372–1379.
[27] K. Seelbach, B. Riebel, W. Hummel, M.-R. Kula, V. I. Tishkov, A. M. Egorov,
C. Wandrey, U. Kragl, Tetrahedron Lett. 1996, 37, 1377–1380.
[28] V. I. Tishkov, V. O. Popov, Biomol. Eng. 2006, 23, 89–110.
[29] K. Hoelsch, I. Sührer, M. Heusel, D. Weuster-Botz, Appl. Microbiol. Bio-
technol. 2013, 97, 2473–2481.
Biotransformations of 2-methylpyrroline were performed in glass
vials with 2 mm of purified enzyme, 8 mm of 2-methylpyrroline, and
10 mm of NAD(P)H. After an incubation of 45 min at 258C, 3-meth-
ylpiperidine was added as internal standard, and the reaction was
stopped with NaOH (final concentration 1.43m). The samples were
extracted with methyl tert-butyl ether (MTBE) and derivatized with
acetic anhydride. Finally, the organic phases were analyzed by gas
chromatography (GC) as described previously (Figures S7 and S8 in
the Supporting Information).[37]
Acknowledgments
The research leading to these results has received support from
the DFG (Deutsche Forschungsgemeinschaft). We want to thank
Prof. Dr. Frances H. Arnold and Dr. Jackson K. B. Cahn for support
regarding the usage of the CSR-SALAD tool.
[30] R. Woodyer, W. A. van der Donk, H. Zhao, Biochemistry 2003, 42, 11604–
11614.
[31] T. W. Johannes, R. D. Woodyer, H. Zhao, Biotechnol. Bioeng. 2007, 96,
18–26.
[32] R. A. Chica, N. Doucet, J. N. Pelletier, Curr. Opin. Biotechnol. 2005, 16,
378–384.
Conflict of interest
The authors declare no conflict of interest.
[33] K. A. Kretz, T. H. Richardson, K. A. Gray, D. E. Robertson, X. Tan, J. M.
Short, Methods Enzymol. 2004, 388, 3–11.
[34] M. Gand, C. Thçle, H. Müller, H. Brundiek, G. Bashiri, M. Hçhne, J. Bio-
technol. 2016, 230, 11–18.
Keywords: activity recovery · cofactor specificity · enzyme
catalysis · imine reductase · site-directed mutagenesis
[1] M. Breuer, K. Ditrich, T. Habicher, B. Hauer, M. Keßeler, R. Stürmer, T. Ze-
linski, Angew. Chem. Int. Ed. 2004, 43, 788–824; Angew. Chem. 2004,
116, 806–843.
[2] R. C. Simon, B. Grischek, F. Zepeck, A. Steinreiber, F. Belaj, W. Kroutil,
Angew. Chem. Int. Ed. 2012, 51, 6713–6716; Angew. Chem. 2012, 124,
6817–6820.
[35] J. K. B. Cahn, C. A. Werlang, A. Baumschlager, S. Brinkmann-Chen, S. L.
Mayo, F. H. Arnold, ACS Synth. Biol. 2017, 6, 326–333.
[36] E. Krieger, G. Koraimann, G. Vriend, Proteins Struct. Funct. Bioinf. 2002,
47, 393–402.
[37] M. Lenz, J. Meisner, L. Quertinmont, S. Lutz, J. Kästner, B. M. Nestl,
ChemBioChem 2017, 18, 253–256.
[3] M. T. Reetz, K. Schimossek, Chimia 1996, 50, 668–669.
[4] M. M. Heberling, B. Wu, S. Bartsch, D. B. Janssen, Curr. Opin. Chem. Biol.
2013, 17, 250–260.
[5] M. J. Abrahamson, E. Vµzquez-Figueroa, N. B. Woodall, J. C. Moore, A. S.
Bommarius, Angew. Chem. Int. Ed. 2012, 51, 3969–3972; Angew. Chem.
2012, 124, 4036–4040.
[6] E. O’Reilly, C. Iglesias, D. Ghislieri, J. Hopwood, J. L. Galman, R. C. Lloyd,
N. J. Turner, Angew. Chem. Int. Ed. 2014, 53, 2447–2450; Angew. Chem.
2014, 126, 2479–2482.
Manuscript received: July 21, 2017
Revised manuscript received: August 22, 2017
Accepted manuscript online: August 28, 2017
Version of record online: November 29, 2017
ChemCatChem 2018, 10, 183 –187
www.chemcatchem.org
187
ꢀ 2018 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim